![Problem setting:-
X : Set of all the instance (eg: set of people)
each described by attributes <age, height>.
C : Target concept the leaner need to learn.
C: X{0,1}
L : Learner have to learn “people who are
skiers”
C(x) =1 : positive training example
C(x) =0 : negative training example
Error of a hypothesis: True error, denoted by errorD
(h), of hypothesis h w.r.t target concept c and
distribution D is probability that h will misclassify an
instance drawn at random according to D.
errorD (h) = pr [c(x) not equal to h(x)]
XED](https://image.slidesharecdn.com/artificialintelligence-220514185946-5caee5fe/85/Artificial-Intelligence-pptx-22-320.jpg)
Artificial Intelligence.pptx
- 1. Nadar saraswathi college of arts & science, theni. Department of cs & it ARTIFICIAL INTELLIGENCE PRESENTED BY G.KAVIYA M.SC(IT) TOPIC:LEARNING FROM OBSERVATION
- 2. LEARNING LEARNING FROM OBSERVATION Forms of learning Ensemble learning Computational learning theory
- 3. LEARNING: Learning is Agent’s percepts should be used for acting. It also used for improving the agents ability to act in the future. Learning takes places as the agents observes, its interactions with the world and its own decision making processes.
- 4. FORMS OF LEARNING: Learning Agent can be thought of as containing a Performance Element, that decides, what actions to take, and a Learning Elements that modifies the performance elements to take better decisions. Three major issues in learning element design Which components the performance element are to be learned. What feedback is available to learn these components. What representation is used for the components.
- 5. Components of Agents are; A direct mapping from conditions on the current state to actions. A means to infer relevant properties of the world from the percept sequence. Information about the way the world evolves and about the results of the possible action the agent can take. Utility information indicating the desirability of world states. Action value information indicating the desirability of action. Goals the describe classes of the state whose achievement maximizes the agent’s utility.
- 6. Classified into three categories: Supervised Learning. Unsupervised Learning. Reinforcement Learning.
- 7. Supervised Learning: The Learning here is performed with the help of teacher. Let us take the example of the learning process of the small child. The child doesn’t know how to read/write. He/she is being taught by the parents at home and by the teacher in school. The children are recognize the alphabet, numerals, etc. Their and every action is supervised by a teacher.
- 8. Continue; Actually, a child works on the basis of the output that he/she has to produce. All these real-time events involve supervised learning methodology. Similarly, in ANNs following the supervised learning, each input vector requires a corresponding target vector, which represents the desired outputs. The input vector along with the target vector is called training pair.
- 9. Continue; In this type of training, a supervisor or teacher is required for error minimization. Hence, the network trained by this method is said to be using supervised training methodology. In supervised learning, It is assumed that the correct “target” output values are known for each input pattern. The input vector is presented to the network, Which result is an output vector. The output vector is the actual output vector. Then the actual output vector is compared with the desired output vector. If there exists a difference between the two output vectors then an error signal is generated by the network. This error signal is used for adjustment of weights until the actual output matches the desired output.
- 10. Un Supervised Learning: The learning here is performed without the help of teacher. Consider, learning process of a tadpole, it learns by itself, that is, a child fish learns to swim by itself, it is not taught by its mother. Thus, Its learning process is independent and is not supervised by a teacher. In ANNs following unsupervised learning, the input vectors of similar type are grouped without the use of.
- 11. Reinforcement Learning: The learning process is similar to supervised learning. In the case of supervised learning the correct target output values are known for each input pattern. But, In some cases, Less information might be available. For example, the network might be told that its actual output is only “50% correct “or so. Thus, Here only critic information is available, not the exact information. The learning information based on the critic information is called reinforcement learning and the feedback sent is called reinforcement signal.
- 12. ENSEMBLE OF LEARNING: Learn multiple alternative definitions of a concept using different training data or different learning algorithms. Combine decisions of multiple definitions, eg. Using weighted voting.
- 13. VALUE OF ENSEMBLES When combing multiple independent and diverse decisions each of which is at least more accurate than random guessing, random errors cancel each other out, correct decisions are reinforcement. Generate a group of base-learners which when combined has higher accuracy. Different learners use different; Algorithm. Hyperparameters. Representations/Modalities/Views. Training sets. Subproblems.
- 14. Ensembles:
- 15. BOOSTING: Also uses voting/averaging but models are weighted according to their performance. Iterative procedure: new models are influenced by performance of previously built ones. * New model is encouraged to become expert for instances classified incorrectly by earlier models. * Intuitive justification: models should be experts that complement each other. There are several variants of this algorithm.
- 16. Continue; STRONG LEARNER: Objective of machine learning. o Take labeled data for training. o Produce a classifier which can be arbitrarily accurate. o Strong learners are very difficult to construct. WEAKER LEARNER: o Take labeled data for training. o Produce a classifier which is more accurate than random guessing. o Constructing weaker learners is relatively easy.
- 17. ADAPTIVE BOOSTING: Each rectangle corresponds to an example, with weight proportional to its height. Crosses corresponds to misclassified examples. Size of decision tree indicates the weight of that classifier in the final ensemble. Using Different Data Distribution * Start with uniform weighting. * During each step of learning. Increase weights of the examples which are not correctly learned by the weak learner. Decrease weights of the examples which are correctly learned by the weak learner.
- 18. Continue; IDEA: focus on difficult example which are not correctly classified in the previous steps WEIGHTED VOTING: construct strong classifier by weighted voting of the weak classifier. IDEA: Better weak classifier gets a larger weight. Iteratively add weak classifiers Increase accuracy of the combined classifier through minimization of a cost function.
- 19. COMPUTATIONAL LEARNING THEORY Computational learning theory characterize. The difficulty of several types of machine learning problem. Capabilities of several types of ML algorithm. CLT seeks answer, question such as; a) “under what conditions is successful learning possible and impossible?” b) “under what conditions is a particular learning algorithm assured of learning successfully?” it means that, what kind of task are learnable, what kind of data is required for learnability.
- 20. Various issues are: Sample complexity: How many training examples are needed for a learner to converge (with high probability) to a successful hypothesis? Computational complexity: How much computational effort is needed for a learner to converge to a successful hypothesis? Mistake bound: How many training examples will the learner misclassify before converging to a successful hypothesis?
- 21. (PAC) Probably Learning an Approximately Correct hypothesis:- A particular setting for the learning problem, called the probably approximately correct(PAC) learning model. This model of learning is based on following points: 1. Specifying problem setting that defines PAC model. 2. How many training examples are required. 3. How much computational are required in order to learn various classes of target functions within PAC
- 22. Problem setting:- X : Set of all the instance (eg: set of people) each described by attributes <age, height>. C : Target concept the leaner need to learn. C: X{0,1} L : Learner have to learn “people who are skiers” C(x) =1 : positive training example C(x) =0 : negative training example Error of a hypothesis: True error, denoted by errorD (h), of hypothesis h w.r.t target concept c and distribution D is probability that h will misclassify an instance drawn at random according to D. errorD (h) = pr [c(x) not equal to h(x)] XED
- 23. PAC Learnability: No of training examples needed to learn a hypothesis h, for which errorD (h) = 0. For these two difficulties, following measures can be taken:- 1. No requirement of zero error hypothesis for learner L. So a bound to error can be set by constant E, that can be made small. 2.Not necessary that learner succeed for every sequence of randomly drawn training example. So learner probably learn a hypothesis that is approximately correct. Bounded by same constant S which is,
- 24. Definition: Consider a concept class define over a set of instance X of kngtn n and a learner L using hypothesis space H. C is PAC learnable by L using H if for all CEC. distribution D over X E such that 0<E<1/2 and S such that 0<S<1/2 Learner L will with Probably at least (1-8) output a hypothesis nEH such that errorD (n) <= E.